In IP telephone, for a terminal to switch a connection destination during communication, a third party or the terminal itself asks a call control server to switch a session and the call control server instructs the terminal to switch a session, thereby establishing a session between each terminal and a new connection destination.
In general, a call control server manages numbers of terminals, so that the number of call controls executed at once is limited. Excess of the number of call controls over a threshold value causes congestion to invite a call loss representing a failure in call control. In a large volume of communication, calling is controlled in some cases for preventing a call loss.
As a mechanism which prevents a call loss, disclosed, for example, in Patent Literature 1 (Japanese Patent Laying-Open No. 2007-251419) as related art is a method of preventing generation of a call loss by abandoning a signal related to call control by a congestion control management unit in a call control server. Abandoning a call control signal denotes that call control is regulated at a stage precedent to actual call control, which is a typical method in telephone call control.
Disclosed in Patent Literature 2 (Japanese Patent Laying-Open No. 2007-243406) as related art is a method, for preventing generation of a call loss caused by a large amount of requests for call control issued by a specific terminal, of instructing a gateway device accommodating the specific terminal to abandon a request for call control which is executed by a call control server. This method is also a method of regulating call control at a stage precedent to actual call control.    Patent Literature 1: Japanese Patent Laying-Open No. 2007-251419    Patent Literature 2: Japanese Patent Laying-Open No. 2007-243406
Related art recited in Patent Literature 1 and Patent Literature 2 have a problem that while generation of a call loss itself can be prevented, in a group conference system, session switching of the group conference system fails.
In simultaneous switching of a session in a group conference system which follows switching of a conference server, when call control is regulated during switching, switching of the entire session fails. In other words, when the above-described related art is applied in simultaneous session switching in the group conference system, while generation of a call loss can be prevented, session switching will fail.
Then, when session switching fails, for complementing the switching failure, it is necessary to complete session switching by redialing by an individual terminal.
Thus, since when targeting only on prevention of call loss generation, in simultaneous switching in a group conference system, terminals simultaneously ask for session switching, so that loads will concentrate on a call control server in a short time period to cause a failure of session switching, as well as sharply increasing loads on the call control server due to generation of redialing following a session switching failure.
First problem is that when sessions are simultaneously switched in a group conference system, loads on a call control server are sharply increased to make a call control system unstable as a whole. The reason is that loads on the call control server are sharply increased in a short time period, so that even if a call control request is abandoned for preventing generation of a call loss, further loads caused by redialing will be generated in the call control server and the loads on the call control server will invite control for preventing a further call loss.